Doomed in din or synced with signal: a study of dissent within Indian monetary policy committee

Abstract

Listen

Translate article

ABSTRACT The paper investigates the effect of latent discord present amongst Indian Monetary Policy Committee members on the accuracy of inflation and growth forecasts. We use FinBERT, a deep learning model, to propose a refined measure of Implicit Dissent and use it to further investigate how such discord can affect the forecast accuracy. Our results show that lower levels of implicit dissent increase uncertainty and impair forecasters’ ability to make accurate predictions. In contrast, higher levels of implicit dissent mitigate uncertainty and help anchor forecasts. Our study concludes that heterogeneity in signals by central bankers increases the mean decision quality of forecasters.